Frequently Asked Questions About SpecLock

SpecLock reads your CLAUDE.md file and converts each rule into a trackable constraint called a "lock." Before any AI-generated code change is applied, SpecLock's semantic detection engine checks for conflicts against these locks. If a proposed change would violate a CLAUDE.md rule, SpecLock blocks it and warns the developer with a clear explanation of which constraint would be broken.

Yes, SpecLock fully supports Cursor alongside Claude Code, Windsurf, Copilot, Cline, and other AI coding tools. SpecLock integrates via the MCP (Model Context Protocol) standard, which means it works with any editor or AI assistant that supports MCP servers. You can also use SpecLock's CLI and pre-commit hooks regardless of your editor.

Install SpecLock instantly with npm: run npx speclock protect in your project directory. This sets up constraint enforcement, pre-commit hooks, and MCP integration in one command. No configuration files are needed — SpecLock automatically discovers your existing CLAUDE.md, .cursorrules, and AGENTS.md rules.

The .cursorrules file is a static text file that provides suggestions to Cursor's AI, but it has no enforcement mechanism — the AI can ignore or forget those rules at any time. SpecLock actively enforces constraints with semantic detection, pre-commit blocking, and conflict checking. Think of .cursorrules as guidelines and SpecLock as laws.

SpecLock provides 51 MCP tools covering the full constraint enforcement lifecycle. These include tools for semantic auditing, patch review, drift detection, lock management, conflict checking, compliance reporting, blast radius analysis, and session continuity. Each tool is purpose-built to prevent AI coding assistants from violating your project's rules.

Yes, SpecLock is completely free and open source under the MIT license. You can install it from npm, use all 51 MCP tools, and deploy it without any cost. The full source code is available on GitHub.

SpecLock's semantic detection engine goes beyond simple keyword matching. It analyzes the intent and meaning of code changes to detect when an AI is trying to violate a constraint, even through euphemisms or indirect modifications. For example, if a lock says "never remove authentication," SpecLock catches attempts to "simplify the login flow" or "refactor auth into a simpler pattern" that would effectively remove it.

SpecLock supports all major AI coding tools including Claude Code, Cursor, Windsurf, GitHub Copilot, Cline, and Bolt.new. It integrates through the MCP protocol standard, CLI commands, and pre-commit hooks, so it works in any development workflow regardless of which AI assistant or editor you use.

To use SpecLock with Claude Code, add it as an MCP server in your Claude Code configuration. SpecLock will automatically load at the start of each session via the speclock_session_briefing tool, providing Claude with your project's constraints, decisions, and history. Claude Code then checks every proposed change against your locks before applying it.

The command npx speclock protect is the one-line setup command for SpecLock. Running it in your project directory initializes constraint enforcement, installs pre-commit hooks, discovers existing rules from CLAUDE.md and .cursorrules, and configures MCP integration. It requires zero configuration and gets SpecLock running in seconds.

SpecLock has 991 tests covering every aspect of the constraint engine — from semantic detection and conflict checking to MCP tool behavior and CLI commands. This extensive test suite ensures reliability and prevents regressions, giving you confidence that SpecLock will correctly enforce your rules in production.

Yes, SpecLock can prevent AI from deleting files. You can create locks that protect specific files, directories, or patterns. SpecLock's pre-commit hooks and patch review tools will block any commit that attempts to delete protected files, and the semantic detection engine catches indirect deletion attempts like moving critical code elsewhere and then removing the original.

The MCP Registry (also known as Smithery) is a public directory of MCP servers that AI coding tools can connect to. SpecLock is listed on the MCP Registry at smithery.ai, making it easy to discover and install. The registry provides a standardized way for developers to find and integrate MCP-compatible tools like SpecLock into their workflow.

SpecLock was created by Sandeep Roy. It is an open-source project available on GitHub and published on npm as the speclock package. Sandeep built SpecLock to solve the critical problem of AI coding assistants forgetting or ignoring project constraints between sessions.